CASE STUDY
From Physical Shelf to Online Storefront: How Miinto Scaled Product Data for Offline Retailors with Dyver
About Miinto and Retailors
Miinto is a leading European fashion marketplace connecting hundreds of retail partners across multiple markets. Their platform enables fashion retailers to sell online by aggregating and distributing product catalogs at scale. Product data quality is core to their business: every product listed must meet strict standards before it goes live.
Retailors is one of Miinto's retail partners, selling exclusively Nike products — both current collections and discontinued lines. With a rich physical inventory but no structured digital product data, they needed a way to get their full catalog online, fast and at scale.
The Challenge
Retailors had no ready-made product data. Everything needed to be sourced, built, and formatted from scratch (text, images, attributes, transformations, and mappings), all structured to meet Miinto's strict platform requirements.
Miinto's feed structure leaves no room for error. Every product must satisfy a series of mandatory fields with precise conditions. The main product image alone follows a defined set of rules. For current Nike products, sourcing is straightforward. For discontinued products, Dyver has implemented a custom search plugin as a solution.
The complexity did not stop at product content. Retailors operates across multiple physical store locations, each with its own pricing in local currency and live stock tracked by RFID. A shoe in stock in Copenhagen at one price may be sold out in Oslo and priced differently in Stockholm. Although they had an enriched catalog, this was only half of the problem. The other half was connecting that static content to a physical reality that shifts store by store, every day, with no digital infrastructure bridging the two sides.
Beyond the technical complexity, there was a bigger structural problem. Miinto's model had always required retail partners to arrive with existing digital infrastructure. That quietly excluded an entire category of potential partners: physical retail chains with strong inventory but no e-commerce data. Miinto knew this segment existed. They had no way to reach it.
The Solution
What Miinto and Retailors Wanted from Dyver
Retailors needed a fully automated pipeline that could handle their Nike catalog end to end. Specifically:
- Source product text and images from Nike's official website
- Convert image formats to JPEG and extract attributes for color, size, brand, and product type
- Map everything to Miinto's taxonomy and generate titles and descriptions that met platform standards
- Select the main image that follows Miinto guidelines
- Update stock and prices every 30 minutes
- Handle discontinued products without breaking the feed when no image exists at the source
For Miinto, the goal went further. They wanted a scalable model they could replicate across other offline retail partners, a way to onboard a type of client they had never been able to work with before.
The Dyver Approach
Dyver built a fully automated enrichment pipeline tailored to Miinto's strict feed requirements. Every mandatory field was accounted for. Image sourcing, format conversion, attribute extraction, and data transformation all ran without human input.
For discontinued products, Dyver implemented exception-handling logic that managed missing images without breaking the feed or flagging items for manual review.
Closing the gap between content and physical reality
Enriching the catalog was only part of the work. The approved product data still had to reflect what was actually on the shelf in each store, in real time.
Dyver pulls live price and stock data from each physical location automatically, twice per hour, and merges it with the enriched catalog. Three rules apply consistently:
- In-stock only. Sold-out products are removed from the feed automatically.
- Local pricing. Each store's feed carries prices in its own currency — EUR, NOK, SEK, or DKK.
- Continuous monitoring. If a feed drops below expected volume, the system flags it before it becomes a problem.
The result: a store-accurate feed for each location, updated continuously, without manual input.
The result was a production-ready, enriched product feed covering multiple store locations across different markets, delivered within weeks of the first test output. More importantly, the pipeline became a replicable template: Miinto now has a working model for onboarding any offline retailer, not just partners who already have data ready.
The Outcome
Before Dyver, Retailors had 47,000 product rows in a spreadsheet with bare minimum product data. No titles. No descriptions. No images formatted for digital. No way to know, from that spreadsheet, what was actually on which shelf in which store, at what price, in what currency. The gap between physical inventory and a publishable online feed was total.
After: a live, per-store feed for every location, enriched and updated twice per hour, showing only what is actually in stock, priced in the local currency of each market. The same product carries a Danish-krone price in Copenhagen and a Norwegian-krone price in Oslo. Sold-out items disappear from the feed automatically. The entire pipeline runs without manual input.
The pilot met its daily order targets from the start. Miinto expanded from the initial feed to multiple store feeds connected directly to their platform. The integration performed well enough that Miinto made a deliberate decision to stop building an internal alternative and move fully to Dyver as a paid solution.
The most significant outcome was not operational, it was strategic. For the first time, Miinto can onboard pure offline retailers as marketplace partners. A business segment that was previously unreachable is now open.
